ELO SYSTEM

[ickoicko]

Seems to me that the "if greater than 600 elo diff, consider it 600 elo diff" meaning 1.2 lowest elo gain (or 1.8 against new players?) could be a problem. However hiddendragon that is mentioned is playing people about 600 elo diff so that would not make a difference. Does the opponent of a new player also get a K value of 60 (50% more) in those games? That could in that case be quite exploitable.

[Doodads]

Something small could be done quickly and would be effective about no elo games.
At the moment if someone doesn't play for months he lose his ranks.
But if he plays with no elo involved his rank is back.
I guess it's a way to say he's back, so his rank too but it seems obvious it should not.
Why being elo ranked if you play with no elo involved ?

[Doodads]

It seems there is a new bug now.
People never lose their rank.
Perfect2050 for example is still caylus rank 6 but his last game was 72 days ago.

Nice to see other things have been fixed.
I guess 500 elo gap could be better than 600 tho.

[Faignantix]

It seems there is a new bug now.
People never lose their rank.
Perfect2050 for example is still caylus rank 6 but his last game was 72 days ago.

Yes, I was wondering the same thing. Is this a bug or did they decide to remove the 60 days rule when they change the rule about the 600 elo difference?

By the way, I agree with Doodads that the 600 gap should be lower.
I have a question about that: is playing a game with a player >600 lower allows to get back the rank or is this considered as a elo off game?
I think the latter sould be preferable to prevent players to only play one game against a 1300 every 2 months to keep their rank...

This period should also be shorter imo, like 30 days.
When you see the number of fake strong players, especially in Caylus, it is obvious that some little things could be improved in the rating system.
The new rules about elo off games and >600 elo gap are nice first steps.

[glitter and doom]

By the way, I agree with Doodads that the 600 gap should be lower.

I'm not sure what do you mean by the elo gap should be lower. The elo gap does not exist any more, it was removed. Previously, in case the difference between two players was more than 600 points, the elo points were calculated as the difference were exactly 600 points. Some strong players were abusing this rule and were playing a lot of games with very weak players (elo diff 800 or 900 points), gaining 1,226 points per victory. Since this limit was completely removed, they will earn much less: 0,699 (700 diff), 0,396 (800 diff), 0,224 (900 diff) and so on. I think this change is good, and I totally support it.

Regarding 60 days rule, I have to disagree with those who think this period should be lower. Sure, if you play 4-5 games, it is not hard to play each game at least once a month. However, if you are competitive in 10+ different games, it is much harder, especially if you play turn-based games only.

Currently, I am ranked in 17 games. I enjoy playing all of them occasionally, but not every month. I think 60 days rule is fine, and it shouldn't be changed (at least not to be decreased).

[Doodads]

Glitter and Doom,

if you enjoy playing all of the games occasionally, but not every month, why should you be ranked ?
It seems obvious to me that someone who doesn't want to play at least 1 game during the month is not supposed to be ranked.

The problem is that you're not playing to become stronger, you're playing to have good stats.
You're avoiding to play each time opponent is better than you, and lot of people on bga do that.
It's your right, but why this way of acting should be rewarded more ?

Is the rank about rewarding fake good players or real good players ?

I agree with Faignatix.

- If someone play against opponent with 500 elo gap or more then no elo involved.

- If someone doesn't play during 30 days then he should lose his rank.

- If he wants to get his rank back then he has to play someone with elo gap smaller than 500 and elo active.

This seems good to me, it will become harder for abusers to be ranked or keep high elo they don't really deserve, and maybe it will encourage them to try to improve instead of huntings bad players for stats.

[glitter and doom]

Doodads, you are playing Caylus only, it is easy for you to play at least one game each month. However, not everyone is like you. To play 20 different games each month is not that easy. And if I haven't played a game for 30 days, it doesn't mean I am not good at it anymore.

It is really silly to say that I avoid players better than me. I don't avoid anyone. I always set up a turn based game without restrictions, which means anyone could join, including experts and masters. And they often do. I've played with masters many times. I played with many No1 and No2 ranked players. And I played with you, 4 times, remember? The last time I played with you, you had much higher rating than me.

[Doodads]

if I haven't played a game for 30 days, it doesn't mean I am not good at it anymore.

if you haven't played a game for 30 days, it means you didn't play for 30 days.
Why should you be ranked at something you don't play?

Some people are risking elo more often than you and you should keep a safe rank without playing ?
You'll have your rank back when you'll play correct opponents with elo on.

Seems totally fair to me.
Plus, being ranked doesn' t mean being good, especially with the elo system failure, there are a lot of examples in top rank players list in lot of games.

I played with many No1 and No2 ranked players

Check your stats, the % of real good players played.

And I played with you, 4 times, remember? The last time I played with you, you had much higher rating than me.

You played me when i was bad player.
Again, 4 times on how many caylus games ? How many time against real good players like ickoicko or tanshu ?
By the way, if you want we can play now

Again don't take it personnaly, i'm not attacking you, you're allowed to play for stats, (which is opposite of playing for improvment and skill).
I just want an elo system that reflects real competition not fake stats.
What is the point of rank system if it doesn't reflect anything ?

[glitter and doom]

You played me when i was bad player.

The last time we have played you had 2148 elo. OK, right, you were a bad player.

[Doodads]

lol.

I told you, the elo system is broken. Someone could have 3000 and still be bad.
I remember our game and yes i was still bad + the point of the thread is not proving you play only bad players.
I said you're not searching for challenge and you re more often going for bad players than good, you could aslo create based turn table inviting real masters.
That doesn't mean you're bad player.


The debate is about a better elo system.

If the goal is to reward real good players (which is suppoed to be when there is a medal called "best player") then the current elo system is broken.
First they should remove all ranks from people who didn't play for 60 days and more.

Then, this seems good :

- If someone play against opponent with 500 elo gap or more then no elo involved.
- If someone doesn't play during 30 days then he should lose his rank.
- If he wants to get his rank back then he has to play someone with elo gap smaller than 500 and elo active.